[sane-devel] scanimage/saned segfault

[Scott Hazen Mueller]

>scott, not enough data from the debug logs there. with both the endianness 
>and alignment issues of gcc on the slug, it could be a host of things. can 
>you try doing some differences in color depth and resolution, and see if 
>it is truly size based, or rather data size based.

OK, here's what I was able to test -

size 8.5x11, gray instead of color - fails
size 8.5x11, resolution 100 instead of 300 - fails
size 6x4, resolution 1200 instead of 300 - works, generates a 45MB .pnm file.

>can you verify that your setup works with an x86 linux box in place of the 
>slug?

I don't have any x86 Linux boxes, but I moved the scanner to my x86 FreeBSD
box.  With the slug as the client and the BSD box as the server, an 8.5x11
scan works.  Prior to that, I did an additional slug-to-slug test, and that
failed.

>can you try openslug?

I probably can, if I swap the scanner to the slug that's not running my VPN...
Does it look like it'd be a valuable test?

Thanks,

                     \scott

>allan
>nslu2-linux moderator, and sane backend maintainer.

>On Mon, 8 Aug 2005, Scott Hazen Mueller wrote:

>> I'm playing with network scanning.  I have a Linksys NSLU2 running Unslung
>> (the opened-up firmware) acting as the scanner server (for an Epson Perfection
>> 1670, backend is snapscan) scanner.  Local scanning works fine at 8.5x11

>> Network scanning works fine at a size of 6x4 (more debug output):
>>
>> However, when I try a network scan at 8.5x11, scanimage dies with a broken
>> pipe and saned segfaults (yet more debug output):
>>